Geoforum 35 (2004) 145–156
www.elsevier.com/locate/geoforum
Linguistic segregation in urban South Africa, 1996
A.J. Christopher
Department of Geography, University of Port Elizabeth, P.O. Box 1600, Port Elizabeth, South Africa
Received 22 November 2001; received in revised form 2 October 2002
Abstract
South Africa is a multi-lingual country with 11 official languages and a recent history where language was frequently used as a
political instrument, notably in the urban areas. Although the cities were initially colonial foundations, as a consequence of rural–
urban migration, the speakers of the various national languages have come into close contact with one another. However, as a result
of the inheritance of apartheid town planning and its emphasis on racial zoning, residential segregation levels between some linguistic groups have been extremely high. An analysis of the 1996 census results reveals that the uniformly high segregation levels
between the speakers of indigenous African languages and the speakers of Afrikaans and English are the direct outcome of
apartheid era town planning. Nevertheless, segregation between the speakers of different African languages may also on occasion be
relatively high where homeland political policies were pursued, although this was the exception rather than the rule. Similarly
segregation between English and Afrikaans speakers was locally high where home language coincided with former racial classification. Few immediate significant changes are anticipated in the present patterns of linguistic segregation, as the inherited apartheid
city structure is proving to be remarkably resistant to transformation.
Ó 2003 Elsevier Ltd. All rights reserved.
Keywords: South Africa; Census; Population; Languages; Urban segregation; Apartheid
1. Introduction
In the second half of the 20th century South African
cities were subjected to apartheid social engineering and
legally enforced racial segregation was pursued with the
object of entrenching White minority rule (Christopher,
2001a; Davies, 1981, 1996; Lemon, 1991; Robinson,
1996; Smith, 1992; Western, 1996). The legacy of this
policy was still powerfully in evidence at the time of
the 1996 population census, which demonstrated the
continuing high levels of racial segregation in the urban
areas of South Africa (Christopher, 2001b). One of the
consequences of apartheid policies was the evolution of
linguistic segregation, where race or population group
and language group were often coincident (van der
Merwe, 1993, 1995). With the repeal of the laws enforcing racially defined residential restrictions, the
imprint of linguistic segregation remains, presenting
significant problems for national integration. The entrenchment of 11 official languages in the South African
E-mail address: [email protected] (A.J. Christopher).
0016-7185/$ - see front matter Ó 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.geoforum.2003.08.007
constitution is indicative of the importance attached to
the issue of linguistic identity by the national government (African National Congress, 1994). It is in the
towns and cities that contact between linguistic groups is
most intense, and although reinforcing apartheid in
some aspects, it also demonstrates a vital aspect of integration and tolerance which is instructive for the
present era of nation-building. This is particularly significant in view of the importance of the politically imposed language usage at schools in sparking the 1976
Soweto student riots (Bonner and Segal, 1998).
It is proposed to examine the degree of residential
segregation between language groups in South African
cities. First, the extent of spatial separation between the
speakers of indigenous and non-indigenous languages is
explored. This distinction closely corresponds with the
most significant legal division within the apartheid city;
namely that between Africans and non-Africans. Secondly, the Afrikaans–English dichotomy is presented,
which on occasion reflected the racial classification of
the population. Thirdly, segregation between the speakers of indigenous African languages is examined, where
segregation was enforced on a piecemeal basis in pursuit
of the grand apartheid programme, centred upon the
146
A.J. Christopher / Geoforum 35 (2004) 145–156
political exclusion of the homelands and their inhabitants from the South African state.
Table 1
Languages of the urban population of South Africa 1996
Language
2. Linguistic patterns
The population of South Africa speaks a wide variety
of languages. Eleven have been granted official status
and are recorded in the census returns (South Africa,
1999a) (Table 1). Two, English and Afrikaans, held
official status prior to 1994 and may be regarded as
European or European-derived languages associated
with the previous White controlling elite. The other nine
are African languages formally systematised from the
speech of the indigenous peoples of the country. In
common with many countries a large number of people
are multi-lingual, using different languages for different
purposes, notably in education, government, commerce,
the home, and socially outside the home (Laitin, 1992).
The exact phraseology of the census questionnaire thus
assumes a high degree of significance in influencing the
response (Arel, 2002). The 1996 census recorded only
Ôfirst home language’, forcing people to choose which
they wished to have recorded against their name, drawn
from a list of the 11 official languages and Ôothers’. It is
notable that the 1996 census was the first to pose a
uniform language question for the entire population
regardless of race. 1
The constant political equation of language with the
concept of the nation, both in Europe and in South
Africa under apartheid, has made the issue of language
identity highly contentious (Williams, 1994). However,
other multi-lingual countries, including India with its
great linguistic diversity, have avoided the destructive
characteristics of linguistic nationalism, through the
promotion of a sense of unity, without the creation of
exclusive ethnic homelands (India, 1955). The government of South Africa is seeking to do the same in the
post-1994 era, through recognition and the creation of
an all-embracing national identity, independent of language. This task is to be supported by the Pan South
African Language Board, which is responsible for the
development of the country’s languages on the basis of
equal treatment and the promotion of multilingualism (Maartens, 1998). The development and promo-
1
It is a measure of official priorities that race has been the subject of
census inquiry since the conduct of the first scientific enumeration in
the Cape Colony in 1865. A question on ability to speak the two
official languages was introduced only in 1918 and then for Whites
only. A question on home language was introduced in 1936, although
only extended to the entire population in 1946. It should be noted that
no Afrikaans or English options were afforded African respondents
nor African languages for the White, Coloured and Asian respondents
(South Africa, 1954).
Afrikaans
English
IsiNdebele
IsiXhosa
IsiZulu
Sepedi
Sesotho
SiSwati
Setswana
Tshivenda
Xitsonga
Other/unspecified
Total
Number of speakers
in urban areas
Urbanised (%)
4,880,923
3,329,501
215,964
3,201,998
4,125,981
1,007,444
2,152,854
283,172
1,508,360
159427
527,878
388,308
84.0
96.2
36.8
44.5
44.8
27.3
65.2
27.9
45.7
18.2
30.1
66.5
21,781,808
53.7
Source: (South Africa, 1999a).
tion of the African indigenous languages has been
viewed as an essential part of the nation-building programme based on cultural diversity (Alexander, 1997).
Indeed it has been suggested that unless such policies are actively pursued ÔThere is a real danger that a
language faultline will displace the racial faultline. . .
to demarcate an unbridgeable gulf between those who
are ‘‘in’’ and those who are ‘‘out’’’ (Alexander, 2002,
p. 96).
Indigenous African languages may be grouped into
three broad Ôsub-families’, Nguni, Sotho and ÔOthers’,
which are closely related members of the South-Eastern
Bantu linguistic family (Greenberg, 1970; Jordan, 1973).
The Nguni sub-family includes IsiXhosa, IsiZulu, SiSwati
and IsiNdebele, which share many broad grammatical
features and word roots. The same applies to the Sotho sub-family of Sesotho, Sepedi and Setswana. Within
the broad sub-families there is a fair degree of mutual
intelligibility. Some differences, such as those between SiSwati and IsiZulu are relatively slight so that
the latter is taught in schools in Swaziland. The ÔOthers’
are Tshivenda and Xitsonga, whose relationships are
more tenuous and belong to linguistic sub-families
whose members reside mainly outside South Africa’s
borders. The close relationship between adjacent language groups, together with regional dialects often
make the precise identification of a spoken language
difficult (Magi, 1990). Indeed it is possible that census enumerators might discern or enter the language expected or politically required, while those subject to
enumeration might offer the answers which they thought
would be most acceptable to the government enumerators. In both cases the results will be unrepresentative
and the inaccuracy virtually impossible to detect.
The two major non-indigenous languages in South
Africa are English and Afrikaans. Afrikaans is derived
from Dutch, although it has evolved, through physical
isolation and the incorporation of a large number of
A.J. Christopher / Geoforum 35 (2004) 145–156
words and grammatical forms from a variety of European, African and Asian languages, into a separate
written and spoken language (Ponelis, 1993). It could
therefore technically be claimed to be indigenous to
Africa. These are the languages of the colonial powers
and their settlers and descendants, although the small
numbers of surviving indigenous Khoisan peoples of
the western part of the country adopted Afrikaans,
leading to the virtual extinction of their own languages
(Crawhall, 1999). Another significant language shift
has been the suppression of Indian languages by English,
and to a far lesser extent Afrikaans, in the Indian
community in the course of the 20th century (Prabhakaran, 1998).
Prior to urbanisation, the indigenous linguistic
groups in South Africa were regionally based in their
distribution (Moseley and Asher, 1994). The urban
areas were largely established in the colonial era and
served the administrative and commercial interests of
the White population. They therefore housed the main
concentrations of English- and Afrikaans-speakers.
However, the urbanisation of the indigenous population
resulted in substantial changes with large-scale migration to the towns, particularly in the late 19th and 20th
centuries. The towns increasingly have come to reflect
the linguistic patterns of the neighbouring rural areas. A
notable exception was the Gauteng industrial region,
which attracted people from all over the sub-continent,
thus creating a multi-lingual urban complex (van der
Merwe and van Niekerk, 1994).
Furthermore, although 88.6% of Afrikaans- and
English-speakers in South Africa were urbanised by
1996, only 42.9% of indigenous language speakers
were resident in the towns. The potential for substantial changes in linguistic ratios in the future is
evident as the urbanisation of the African population
proceeds and the African proportion of the urban
population increases. This may be gauged from the
position of the largest linguistic group, the IsiZuluspeaking population. In 1970 when only 30.9% of
IsiZulu-speakers resided in urban areas, they comprised
12.1% of the urban population (South Africa, 1973). As
a result of rural–urban migration, in 1996 some 44.8%
resided in the formal urban areas when they constituted
18.9% of the urban population. 2 The ratio of indigenous to non-indigenous language speakers was further
modified as a consequence of the reduction of the White
population between 1991 and 1996. The combined
urban total of English and Afrikaans speakers thus
remained constant at approximately 8.2 million, while
2
The census definition of an urban area was dependent upon its
administrative status. Thus considerable numbers of people who were
functionally part of the urban areas, but lived in nominally rural areas,
have been excluded from this discussion.
147
the speakers of indigenous languages increased by approximately 10% in the same period (South Africa,
1992). 3
The cities thus became areas of linguistic complexity
and also of rapid evolution and assimilation (Calteaux,
1996). The international and national dominance of
English in the fields of business, higher education and
administration has been reflected in the widespread
adoption of English in primary, secondary and tertiary
education in Anglophone Africa (Mazrui, 2002). The
result has run counter to South African official language
policies as
It would seem that modernisation in South
Africa, and the inexorable urbanisation in particular, is undermining the possibilities for the first
alternative (that is promoting additive bilingualism)
and that the more realistic option is a straight for
English approach except in linguistically homogeneous classes where there is little exposure to
English outside the classroom or where parents expressly request an alternative (South Africa, 1999b,
p. 23).
So far this process has not produced a significant
language shift at household level, although the possibility exists and is being resisted in official circles
through the active promotion of indigenous languages
at school (Brown, 1998; McDermott, 1998).
In an age when the racial classification system
has been rejected by many census authorities, home
language remains a significant social and cultural
marker (Mohanty and Momin, 1996). Indeed language
may be a key signifier in South Africa for the concept of
Ôorigins’, which has been identified as more meaningful for the popular understanding of American diversity,
than Ôrace’ in the United States census (Hirschman et al.,
2000). The same may become true in post-apartheid
South Africa. Segregation between linguistic groups
may therefore provide a measure of national integration parallel with the (still) required racial classification. In order to assess the significance of linguistic
separation in the urban areas, it is proposed to examine
the degree of segregation between the speakers of
indigenous and non-indigenous languages, and then
3
South African census statistics are subject to severe limitations.
Between 1976 and 1994 the Ôindependent’ homelands of Bophuthatswana, Ciskei, Transkei and Venda were responsible for the collection
of data, including the conduct of censuses. The results were conflicting
and the household questionnaires differed from those in South Africa.
In 1991, Bophuthatswana (1991), Transkei (1994) and Venda (1993)
conducted their own censuses. The disturbed political situation in
Ciskei (1988) prevented the conduct of a census, and reliance must be
placed upon estimates and projections.
148
A.J. Christopher / Geoforum 35 (2004) 145–156
the extent of segregation within these two broad
groupings.
3. Measuring segregation
The standard index of dissimilarity as devised by
Duncan and Duncan (1955) has been adopted to
distinguish spatial segregation levels between linguistic
groups. The index’s ease of calculation and broad
comparability have resulted in its widespread use, despite its limitations (Brun and Rhein, 1994; Christopher,
1992a,b; Frey and Farley, 1996; Massey and Denton, 1988; Telles, 1992; White, 1986). The index is
calculated by comparing the spatial distribution of
two population groups within a town. The index of
dissimilarity is calculated according to the following
formula:
IDxy ¼ 0:5Rjxi yi j
where xi represents the percentage of the X population
in the ith census tract, yi represents the percentage of the
Y population in the ith census tract, and IDxy represents
the index of dissimilarity between the spatial distribution of the X and Y populations.
The index of dissimilarity is measured on a scale from
0 (identical distribution) to 100 (completely segregated).
In the interpretation of the results, Kantrowitz (1969)
suggested that indices of under 30 represented completely random numbers and therefore populations
which could be regarded as integrated, while indices
above 70 represented a recognisable degree of structural
or legally enforced segregation. Values between 30 and
70 may indicate voluntary Ôsocial’ segregation between
two identifiable groups without the intervention of legalised coercion. It is an illustration of the impact of
racially defined apartheid that the median index of dissimilarity between Africans and Whites in South African
towns increased from 78 in 1951 to 95 in 1991. This level
represents Ôhyper-segregation’ resulting from highly coercive legal enforcement, from which South Africa is
now emerging (Massey and Denton, 1993). Indeed, in
the present discussion, it is proposed that an index value
of 90 or over might be recognised as an indicator of
Ôhyper-segregation’.
Census returns at enumeration tract level for the
urban areas of the country were acquired from Statistics
South Africa, which conducted the national population
census in October 1996. The urban census enumeration
tracts, on average, housed 481 people. The statistics
were analysed for the urban areas on the basis of
the magisterial districts, except where metropolitan or
physical urban areas were divided between two or more
districts. Thus district statistics for the larger metropolitan areas such as Johannesburg, Pretoria, Durban
and Cape Town were grouped together for purposes of
calculation. In addition a number of centres such as
Pietersburg and Sheshego (Northern Province) and
King Williams Town and Zwelitsha (Eastern Cape) were
reunited. 4 In these cases district boundaries had been
drawn by the former government to excise the African
suburbs from the remainder of the town, in order to
place them under the administrative and political control of the former homeland governments. However,
provincial boundaries have been respected in the analysis. Thus the population displaced from Pretoria to
neighbouring Bophuthatswana has been included in the
North West Province, not Gauteng (Gervais-Lambony
and Guillaume, 1999).
Some numerical constraints were introduced into the
calculations. Towns with under 5000 inhabitants were
excluded from the investigations. Similarly language
groups with less than 100 members or one percent of the
population, whichever was the greater, were also excluded. In common with other studies, it was considered
that the inclusion of small towns and communities might
have exerted an erratic bias on the findings (Farley and
Frey, 1994). A total of 249 separate urban towns and
cities were therefore included in the study.
3.1. Indigenous/non-indigenous
African residential locations or townships had been
established by the government since colonial times and
residence in them legally prescribed prior to the enactment of apartheid laws in the 1950s. Thus in many
urban areas the segregation of indigenous Africans from
the remainder of the population had been official policy
for 100–150 years prior to the repeal of such discriminatory laws in 1991. Even the numbers of resident African domestic workers in the designated White areas
were strictly controlled (Mather, 1987). However,
there were occasional survivals, where integrated communities were Ôforgotten’ by the town planners, but the
numbers involved were very small (Maharaj, 1999). By
1991 only approximately 7% of the urban population of
South Africa lived outside their designated racially defined zones, mostly in the White group areas as domestic
servants and workers. Thus the legal enforcement of
residential segregation, which was of fundamental importance in the evolution of South African cities,
resulted in the effective spatial separation between the
members of the indigenous and non-indigenous language groups, although that was not the specific intent
of the policies.
4
Northern Province was subsequently (2002) renamed Limpopo
Province. The names of some of the metropolitan areas and municipalities were also changed in 2002. The original nomenclature used in
the census has been retained.
A.J. Christopher / Geoforum 35 (2004) 145–156
Furthermore, in spite of the long history of interpersonal contact between members of the two communities, linguistic assimilation was comparatively rare.
Thus few Africans adopted English or Afrikaans as their
home languages, notwithstanding their status as official
languages and their utility in business and government. Similarly few Coloured, Indian or White families
adopted the use of indigenous languages in the home.
Consequently in the urban areas only 0.2% of the urban
Coloured, Indian and White populations returned an
indigenous language as Ôfirst home language’, while only
2.0% of urban Africans recorded English or Afrikaans
as theirs (South Africa, 1999a). 5 Such a high coincidence of language with racial or population group
means that this index is thus a measure of the effectiveness of apartheid planning and its continuing legacy
in the post-apartheid era. As might be anticipated, linguistic segregation between indigenous and non-indigenous language speakers was of a similar level to that
recorded between Africans and non-Africans. The development of the dominant language of public record,
English, as a home-based lingua franca among the
population as a whole is not imminent.
It should be noted that there were some 35 urban
areas with populations over 5000 inhabitants, which
lacked either an indigenous or non-indigenous language
speaking group of more than 100 people or one percent
of the total. These towns were situated either in one of
the three Cape provinces, where African language
speakers were often present only in small numbers, or in
the former African homelands where English and Afrikaans speakers were similarly not present in significant
numbers. In the majority of cases such towns were effectively mono-lingual. Thus in the following section a
total of 214 urban areas were included in the analysis
(Fig. 1).
With few exceptions the indices recorded were exceptionally high, indicating the extensive occurrence of
hypersegregation (Table 2). The national median index
was 91 and three-quarters of all values were contained
within the range 65–97. None of the indices were below
30, and so ascribable to purely random numbers and
integrated populations. Furthermore, the degree of segregation was fairly uniform across the country with little
regional variation. Only in the Free State was the level
of dissimilarity between the two groups statistically
significantly higher than in the other provinces. 6 This
reflects the inherited legacy of systematic segregation in
that province, dating back to the 19th century, which
was a feature of previous censuses throughout the 20th
149
Fig. 1. Indices of dissimilarity between indigenous and non-indigenous
language speakers.
Table 2
Median provincial indices of dissimilarity by language groups 1996
Province
Indigenous/
non-indigenous
Afrikaans/
English
Largest indigenous/second
indigenous
Western Cape
Eastern Cape
Northern Cape
Free State
KwaZulu–Natal
North West
Gauteng
Mpumalanga
Northern
SOUTH AFRICA
91
91
79
94
91
91
91
92
92
91
62
63
–a
26
49
45
39
38
43
51
–a
–a
–a
25
66
41
38
40
40
37
Source: Calculated by the author from statistics supplied by Statistics
South Africa.
a
Under five towns.
century (Christopher, 1990, 1994). It should also be
noted that there was no significant relationship between segregation level and the size of the town. 7 The
administrations of small towns had been just as effective in imposing the apartheid city model upon their
populations as those in the large cities. However,
segregation levels did increase significantly with the
ratio between indigenous and non-indigenous language
speakers. Thus the higher the proportion of indigenous language speakers, the higher the index of dissimilarity.
5
Some 2.0% of the urban African population was born outside
South Africa.
6
The Mann–Whitney U test for significance was adopted for
comparisons between the index levels in different provinces.
7
Spearman’s rank correlation coefficient was adopted for these
calculations.
150
A.J. Christopher / Geoforum 35 (2004) 145–156
The high dissimilarity index values between the indigenous and non-indigenous language speakers are indicative of the continuing spatial separation between
Africans and other population groups and the absence of
a significant community language shift. The consequences of post-apartheid political developments have
had comparatively little impact upon the basic structure
of South African towns and cities. Limited integration
has been noted as mortgage finance was extended to
African families wishing to purchase property in the
former White areas (Kotze and Donaldson, 1998). In
addition the central areas, notably the high rise apartment blocks, in a number of towns and cities have been
occupied by Africans, but with the effect of converting a
formerly White group area into an African ghetto or
ethnic village (Guillaume, 2001). Land invasions and
resettlements, although disrupting the broad apartheid
sectoral racial zoning plan, have usually produced highly
segregated settlements (Gigaba and Maharaj, 1996;
Oelofse and Dodson, 1997; Saff, 1996). Furthermore,
formally planned housing schemes have had little different result, perpetuating racially segregated suburbs
(Bremner, 2000). There have been exceptions, notably
on the Cape Flats (Oldfield, 2000). Land restitution
programmes have been limited in scale and have not
succeeded in reestablishing pre-apartheid integrated
communities (Popke, 2000). The vibrancy and vitality of
African township cultures similarly suggest that there is
no concerted desire to abandon them, even if the extreme
poverty of the majority of the inhabitants were to be
alleviated (Houssay-Holzschuch, 1999). As a result there
appears to be a sustained programme to rehabilitate and
upgrade living conditions in the townships, with all the
implications for continued segregation that this implies
(Harrison et al., 1997). In similar vein, the construction
of the gated White suburb, with its walls, electrified
fences and security guards, represented the most visible
defence of the apartheid inheritance and the maintenance
of the status quo (Hook and Vrdoljak, 2002).
After the election of the new democratic, but
economically non-interventionist, government in 1994,
there was no concerted, government-driven, programme
aimed at racial and ethnic integration and the deliberate
construction of a post-apartheid city (Bond, 2000). This
maybe parallels the political debate over the desirability
of cultural and linguistic assimilation as opposed to
social pluralism (Ratcliffe, 1996, p. 301). Indeed the
overwhelming poverty of the African population is
likely to inhibit integration, producing cities segregated
by income level, which closely reflected language and
ethnicity as identified in Mexico and Brazil (de Fuentes
and Perez Medina, 1998; Telles, 1995). Such an outcome
may be liable to some modification in the face of White
emigration, particularly if it were to become similar in
scale to that experienced by Zimbabwe after 1980 or
Algeria in 1962 (Cumming, 1993; Eichler, 1977).
3.2. Afrikaans/English
Segregation between English and Afrikaans speakers
was never formalised under apartheid as the two
languages were spoken predominantly by the White,
Coloured and Indian population groups, who were, however, segregated from one another under the Group
Areas Act (1950). It should be noted that the three
groups differed substantially in their linguistic composition, with 78.9% of urban Coloured people returning
Afrikaans as their first home language, but only 1.4% of
urban Indians doing so in 1996 (South Africa, 1999a).
The White population was more evenly balanced
with 55.9% returned as speaking Afrikaans and 40.4%
speaking English. An earlier study had demonstrated
relatively low levels of segregation between the two
language groups within the White community (van
Bergen and Olivier, 1983).
Marked regional variations in distribution of the two
groups were evident so that only in 147 towns was it
possible to calculate an index between them (Fig. 2). A
total of 102 towns were therefore excluded as the numbers of one or other group were either under 100 or one
percent of the total population. Afrikaans was dominant
among non-indigenous language users in most urban
areas outside KwaZulu–Natal, where English dominated to the virtual exclusion of the other language.
Segregation levels were substantially lower than those
between indigenous and non-indigenous language speakers. The national median index value was only 51.
Indeed in 24 urban areas the index was below 30, indicating random numbers and hence effective integration
between the two groups. However, some 19 towns recorded indices above 70, indicating a high degree of
Fig. 2. Indices of dissimilarity between English and Afrikaans speakers.
A.J. Christopher / Geoforum 35 (2004) 145–156
structural segregation. In these cases the continuing effects of the previously legally enforced segregation between the Coloured, Indian and White populations was
in evidence, where the dominant languages within these
groups were different. For example, in the Albany district (Eastern Cape) the disparity between the two
groups was particularly marked and may be ascribed to
the different language profiles of the three population
groups. Thus although 98% of non-indigenous language
speakers in the former Coloured group areas spoke
Afrikaans, only 23% of non-indigenous language speakers
in the former White and Indian group areas did so. The
result was a high degree of segregation between the two
language groups as a direct result of the coincidence of
race and language within a system of racially based
segregation. Although the overall index of dissimilarity
between Afrikaans and English speakers in the Albany
district was at coercive levels (78), within the former
White and Indian group areas, levels were substantially
lower (58 and 52 respectively).
Furthermore, there was no significant difference between segregation levels according to size of urban area
or size of the combined Afrikaans and English speaking
groups. However, there was a significant relationship
between the relative sizes of the two groups, both to one
another and in relationship to the remainder of the
population. Thus, the more evenly balanced the English
and Afrikaans speaking groups, the greater the degree of
integration, while increasing dominance by one group
resulted in increasing levels of segregation. In addition
there was a significant positive relationship between the
combined percentage of Afrikaans and English speakers
in the total urban population and the level of segregation.
The legacy of the apartheid city was well illustrated in
Port Elizabeth, where the residential distribution of the
population in 1996 still conformed closely to the former
racial zoning patterns (Fig. 3). The extremely high index
of dissimilarity (96) between the indigenous, predominantly IsiXhosa-speaking, population and the remainder was immediately apparent. This reflected the long
history of African segregation in the city and the spatial
coincidence of IsiXhosa-speaking population with the
former African zoned suburbs. Within the former
White, Coloured and Indian suburbs, there was a more
complex pattern, reflecting the dominant position of
English among the Indian, and to a lesser extent the
White, populations and Afrikaans among the Coloured
population. Thus the index of dissimilarity value in Port
Elizabeth as a whole between Afrikaans and English
speakers was calculated to be 61. However, within the
former Indian and White group areas the indices were
substantially lower (Table 3). The contrasting higher
levels in the former Coloured group area were attributable in part to differential levels of education and income between the two language groups.
151
Fig. 3. Distribution of IsiXhosa, English and Afrikaans speakers, Port
Elizabeth.
Table 3
English and Afrikaans speaking population of Port Elizabeth
Former group
area
Afrikaans
speakers
English
speakers
Index of
dissimilarity
Coloured
Indian
White
Port Elizabeth
148,163
378
65,486
217,485
24,141
6567
76,314
108,356
61
29
37
61
Source: Statistics South Africa, Population Census 1996, Language
tables: Port Elizabeth.
3.3. African language groups
In only 136 urban areas was it possible to calculate
indices for the different distributions of the two largest
African language groups, where both exceeded one
percent of the total population (Fig. 4). Thus 113 towns
were excluded, mainly for lack of a numerically significant second indigenous language group. It should be
noted that the majority of the towns in the coastal
152
A.J. Christopher / Geoforum 35 (2004) 145–156
Fig. 4. Indices of dissimilarity between the two largest indigenous
linguistic groups.
provinces were excluded, as either IsiXhosa or IsiZulu
was so dominant that the members of no other African
language group exceeded one percent of the population.
This phenomenon is illustrated by the Cape Town metropolis where the census recorded 543 190 IsiXhosa
speakers, while the next largest group, Sesotho, recorded
only 3587 speakers.
Again segregation levels were comparatively low,
with a national median value of 37. Indeed some 32.4%
of towns recorded indices below 30, indicating the
purely random distributions of integrated groups. The
indices of only four towns were at coercive levels, above
70, and these appear to be the result of small concentrated numbers of the second language group, related to
the presence of long-distance migratory labourers
housed in segregated hostels. Regionally, the Free State
recorded significantly lower levels of segregation than
the rest of the country, while KwaZulu–Natal recorded
significantly higher levels. The relatively cosmopolitan
character of the African urban areas in the Free State
contrasted to the virtual mono-linguistic makeup of
those in KwaZulu–Natal. In addition there was no sig-
nificant difference in segregation levels between the
former homeland towns and those situated outside the
homelands.
The analysis of other variables indicated that little
pattern could be distinguished in other potential influences, as investigated in the earlier discussion of the
indigenous to non-indigenous indices of dissimilarity.
There were no significant relationships between level
of segregation and size of total population, number of
indigenous language speakers or the proportion of
the population speaking indigenous languages.
The high degree of uniformity is all the more surprising when it is remembered that African ethnolinguistic segregation was one of the objectives of the
National Party government, in furthering mother tongue education and in the political linkage policies between the homelands and the urban areas (Christopher,
1989). In the African townships in Gauteng, in particular, areas were zoned for individual language groups or
at least for separate areas for the Nguni, Sotho and
ÔOthers’ linguistic families after 1954 (Morris, 1980).
However, there was a chronic housing shortage in the
African townships and hence the process of housing
allocation was usually under severe pressure and the
language policy could not be consistently applied. Indeed the general bureaucratic approach of adhering to
the waiting list order in the allocation of municipal
housing tended to favour the dispersal of linguistic
groups. Furthermore, there were no resources available
for the policy to be applied to those housing areas already occupied prior to the introduction of linguistic
zoning. Thus in the country as a whole, there was no
significant difference between the indices of dissimilarity
of the speakers of languages from different linguistic
Ôfamilies’ and those belonging to the same families.
However, if Soweto (Johannesburg) is examined, the
indices of dissimilarity between the various indigenous
linguistic groups exhibit a limited degree of success for
the zoning policy. Thus segregation levels between the
speakers of languages in different linguistic families were
consistently higher than those between the speakers of
languages within the same family (Table 4). Nevertheless, the two largest groups, IsiZulu- and Sesotho-
Table 4
Indices of dissimilarity between African language groups in Soweto 1996 (above the diagonal) and 1985 (below the diagonal)
IsiXhosa
IsiZulu
Sepedi
Sesotho
Setswana
Tshivenda
Xitsonga
Population 1996
Population 1985
IsiXhosa
IsiZulu
Sepedi
Sesotho
Setswana
Tshivenda
Xitsonga
–
22
66
62
69
74
68
78,137
71,312
33
–
67
63
68
73
67
365,258
223,469
67
66
–
32
24
75
71
56,151
60,584
58
57
36
–
29
79
74
157,352
103,059
68
66
32
36
–
80
75
125,373
118,967
73
72
74
72
76
–
32
31,387
26,404
61
63
69
64
71
42
–
70,250
49,932
Source: Statistics South Africa, Population Census 1996, Language tables: Soweto, and tables supplied for 1985 census.
A.J. Christopher / Geoforum 35 (2004) 145–156
speakers, registered only a moderate index of dissimilarity (57). There would appear to be no linguistic
evidence that the potential ethnic polarisation associated with the Inkatha Freedom Party-African National
Congress murderous conflict in the period from 1984 to
1994 was translated into spatial separation (Taylor and
Shaw, 1994). The emergence of homogeneous ethnic
residential security zones, noted elsewhere on the continent during periods of internal civil conflict, did not
take place in South Africa (Pourtier, 2000). The two
main linguistically identified groups of protagonists
in the conflict remained residentially integrated with
the IsiXhosa–IsiZulu index (33) barely above that accountable by purely random numbers. This finding
would suggest that the conflict remained essentially
political rather than ethnic in character.
An examination of the population of Soweto also indicated the increasing dominance of the major languages.
Between 1985 and 1996 there was a recorded increase of
27.3% in the total population of Soweto (South Africa,
1985). However, the two largest groups recorded increases in excess of the average. Thus IsiZulu-speakers
registered a 63.4% increase and Sesotho-speakers a 32.3%
increase. 8 In contrast IsiNdebele-speakers recorded a
73.8% decline and SiSwati-speakers an 84.0% decline,
indicating a possible linguistic identification, even assimilation, with IsiZulu, following the demise of the
homeland policy, which had promoted the other two
languages for political purposes. Increased levels of literacy and multilingualism, together with a decline in
linguistic nationalism, have led to an apparent fluidity
of the boundaries between related languages, generally
to the benefit of the larger grouping (Msimang, 1998).
An important indicator in this respect is the high degree
of language mixture in general conversation by the
speakers of the minor languages (South Africa, 2000).
Otherwise the relative stability of segregation levels may
be gauged from a comparison of the 1985 and 1996
indices (Table 4).
In contrast to Soweto, Daveyton in Benoni was
constructed and occupied, from its inception, according
to ethno-linguistic zoning principles (de Swardt, 1970).
Zones were provided for most of the recognised linguistic groups in the region and a concerted effort was
made to allocate houses according to the master plan
(Fig. 5). However, the compilers of the zoning plan
could not have anticipated the apparent weak demo-
8
Although the census in 1985 sought to ascertain a person’s mother
tongue, the tables were presented in Ônational units’, with the
implication that it was ultimately a political rather than linguistic
affiliation which was sought. This has substantial implications in the
case of the Swazi and Ndebele Ônational units’, where the daily
common speech was often closer to IsiZulu. Changes in the phrasing of
the language question in the census make inter-census comparisons
problematical (Broeder and Extra, 1999; Maartens, 1998).
153
Fig. 5. Former linguistic zoning and dominant language group in
Daveyton.
graphic representation of the speakers of IsiNdebele,
SiSwati and Tshivenda, nor was provision made for the
future dominance of IsiZulu-speakers. Thus the three
zones set aside for the former groups became effectively
occupied by IsiZulu-speakers. In addition the Xitsonga
zone survived but was heavily penetrated by IsiZuluand IsiXhosa-speakers. In contrast the Setswanaspeakers were not provided with a zone, but were
housed with Sepedi-speakers. As a result they occasionally constituted the dominant group in the Sepedi
zone, with a markedly low inter-group index of dissimilarity (35) between them. The comparative success of
enforced linguistic zoning may be gauged from the key
IsiZulu–Sepedi index of dissimilarity which stood at 73,
while the Xitsonga–Sepedi index measured 78. Both
indices indicated a coercive measure of segregation,
which was the aim of the government planners (Table 5).
Even the Sepedi–Sesotho index measured 58, despite the
linguistic affinity of the two groups. The equivalent (57)
IsiZulu–Xitsonga index is a measure of the former
speakers’ penetration of the latter’s zone, as a result of
the chronic housing shortage.
154
A.J. Christopher / Geoforum 35 (2004) 145–156
Table 5
Inter-group indices of dissimilarity in Daveyton (above diagonal) and the remainder of Benoni (below the diagonal), 1996
IsiNdebele
IsiXhosa
IsiZulu
Sepedi
Sesotho
Setswana
SiSwati
Xitsonga
IsiNdebele
IsiXhosa
IsiZulu
Sepedi
Sesotho
Setswana
SiSwati
Xitsonga
–
44
35
44
47
57
40
37
84
–
34
32
34
47
35
40
53
74
–
30
30
42
36
41
75
82
73
–
33
42
38
40
76
77
70
58
–
39
42
45
81
85
78
35
51
–
54
50
50
75
48
73
72
79
–
41
77
71
57
78
73
81
65
–
Total Daveyton
Total remainder of
Benoni
3560
9847
11,931
18,832
39,621
91,376
13,318
26,405
9321
19,620
6897
6286
1723
6203
6192
15,437
Source: Statistics South Africa, Population Census 1996, Language tables, Benoni.
The limited extent of African linguistic segregation in
the remainder of Benoni offers a significant contrast to
patterns resulting from the zoning of Daveyton. On
average inter-group indices of dissimilarity were 28
points lower elsewhere in Benoni. The Tshivenda indices
have been excluded from consideration as Tshivendaspeakers constituted under one percent of the population in both areas. It is notable that whereas in Daveyton,
virtually all indices were at coercive levels, above
70, none of those in the remainder of Benoni exceeded
57. Most noticeable was the IsiZulu–Sepedi index which
measured 30 in the remainder of Benoni, a value akin to
a random distribution, but had been at coercive levels in
Daveyton. Furthermore, there was no significant difference between the segregation levels of the speakers of
languages in different linguistic Ôfamilies’, when compared with speakers from the same linguistic family in
the area of Benoni outside Daveyton. These findings
suggest that segregation between the speakers of different African languages, where government intervention
had not been coercive in the past, was insignificant, with
no evidence of linguistic enclaves in either the preapartheid township (Wattville) or the more recent informal settlements surrounding Daveyton.
4. Conclusion
Although the urban areas are often regarded as cultural and linguistic melting-pots, most South African
cities remain essential mono-, bi-, or at most tri-lingual.
The majority of towns and cities have drawn both their
indigenous and non-indigenous language speakers from
the immediate rural hinterland thereby reinforcing this
characteristic. Only in the industrial heartland of
Gauteng, which has attracted large numbers of people
from all over the country, and indeed from beyond its
borders, can speakers of all 11 official languages be
found in sufficient numbers to suggest the emergence of
a fully multi-lingual urban region. Within such an en-
vironment there is a pressing need for a lingua franca.
The dominance of English in the conduct of government
and commerce has placed pressure for education to be
conducted in that language for the purpose. The danger
which this poses for indigenous languages has evoked a
significant state response in the promotion of all 11 official languages, although there is evidence that the
smaller linguistic groups are being assimilated into the
larger. State resistance to the dominance of English, for
the moment, suggests that a major language shift, as
experienced by the Indian population of the country in
the 20th century, is not imminent.
The observed patterns of residential linguistic segregation in South African cities are dominated by the
fundamental indigenous African–non-African cleavage
in national society. The division constitutes a tangible
inheritance from the apartheid and earlier eras and
segregation between the two sectors is consequently
maintained at remarkably high levels and indeed may be
described as hypersegregation. This linguistic divide may
remain as marked as the racial divide which it paralleled
as South African cities undergo the slow process of
democratic transformation. Anticipated changes will
involve, first, the partial undoing of the apartheid spatial
segregation plan and, secondly, the accommodation of
massive expansion as the urbanisation of the indigenous
population accelerates, freed from earlier controls on
urban residency. Nevertheless, post-apartheid reconstruction and development programmes have tended to
re-enforce the lines drawn in the previous era. Owing to
the extreme poverty of the vast majority of the African
population the basic racial and linguistic divide is unlikely to be eroded rapidly, without a radical government-driven urban restructuring programme. The
absence of English home language speakers within such
areas further reduces the possibility of significant language shift in this respect.
Residential segregation between English and Afrikaans speaking people shows greater potential for reduction as the Coloured, Indian and White populations
A.J. Christopher / Geoforum 35 (2004) 145–156
return to the more integrated pre-apartheid patterns,
based on economic and social status. Segregation levels
between the speakers of the various indigenous languages are substantially lower and the localised inheritance of legally enforced separation will almost certainly
decline as the housing market becomes more flexible. It
may be concluded that where legal coercion has not
been exercised in the past, linguistic segregation is slight
and differences in indigenous home languages are clearly
of little importance in determining African residential
patterns within the city. This shared urban community
experience is of significance in fostering a sense of national cohesion. In view of the overwhelming and increasing African majority in the urban areas of South
Africa this is the remarkable change from the segregationist past without a common national identity. However, language will remain a significant racial, social and
economic indicator, when the broad division between
the speakers of indigenous and non-indigenous languages is considered.
Acknowledgements
The Financial assistance of the Division for Social
Sciences and Humanities of the National Research
Foundation (South Africa) towards this research is
hereby acknowledged. Opinions expressed and conclusions arrived at are those of the author and are not
necessarily to be attributed to the National Research
Foundation. The author wishes to thank Wilma Britz
for the preparation of the diagrams.
References
African National Congress, 1994. The Reconstruction and Development Programme. Umanyano Publishers, Johannesburg.
Alexander, N., 1997. Language policy and planning in the new South
Africa. African Sociological Review 1, 82–89.
Alexander, N., 2002. An Ordinary Country: Issues in the Transformation from Apartheid to Democracy in South Africa. University
of Natal Press, Pietermaritzburg.
Arel, D., 2002. Language categories in censuses: backward- or
forward-looking? In: Kertzer, D.I., Arel, D. (Eds.), Census and
Identity: The Politics of Race, Ethnicity and Language in National
Censuses. Cambridge University Press, Cambridge, pp. 92–120.
Bond, P., 2000. Elite Transition: From Apartheid to Neoliberalism in
South Africa. Pluto, London.
Bonner, P., Segal, L., 1998. Soweto: A History. Maskew Miller
Longman, Cape Town.
Bophuthatswana, 1991. 1991 Population Census, vol. 1, Part 1.
Statistics Branch, Department of Economic Energy Affairs, Mines
and Planning, Mmabatho.
Bremner, L.J., 2000. Post-apartheid urban geography: a case study of
Greater Johannesburg’s Rapid Land Development Programme.
Development Southern Africa 17, 87–104.
Broeder, P., Extra, G., 1999. Language, ethnicity and education.
Multilingual Matters, Clevedon.
155
Brown, D., 1998. Educational Policy and the Choice of Language in
Linguistically Complex South African Schools Education Policy
Unit. University of Natal, Durban.
Brun, J., Rhein, C., 1994. La segregation dans la ville. L’Harmattan,
Paris.
Calteaux, K., 1996. Standard and Non-standard African Language
Varieties in the Urban Areas of South Africa. Human Sciences
Research Council Publications, Pretoria.
Christopher, A.J., 1989. Apartheid within apartheid: an assessment of
official intra-black segregation on the Witwatersrand, South
Africa. Professional Geographer 41, 328–336.
Christopher, A.J., 1990. Apartheid and urban segregation levels in
South Africa. Urban Studies 27, 421–440.
Christopher, A.J., 1992a. Urban segregation levels in the British
Overseas Empire and its successors in the twentieth century.
Transactions of the Institute of British Geographers 17, 95–107.
Christopher, A.J., 1992b. Ethnicity, community and the census in
Mauritius 1830–1990. Geographical Journal 74, 28–33.
Christopher, A.J., 1994. Segregation levels in the late-apartheid city
1985–1991. Tijdschrift voor Economische en Sociale Geografie 85,
15–24.
Christopher, A.J., 2001a. The Atlas of Changing South Africa.
Routledge, New York.
Christopher, A.J., 2001b. Urban segregation in post-apartheid South
Africa. Urban Studies 38, 449–466.
Ciskei, 1988. Statistical Abstract of Ciskei 1988. Ciskei Central
Statistical Service, Bisho.
Crawhall, N., 1999. San and Khoe rights, identity and language
survival in South Africa. In: Maharaj, G. (Ed.), Between Unity and
Diversity. Essays on Nation-Building in Post-Apartheid South
Africa. David Philip, Cape Town, pp. 33–75.
Cumming, S.D., 1993. Post-colonial urban change in Harare: a case
study. In: Zimyama, L.M., Trevera, D.S., Cumming, S.D. (Eds.),
Harare: The Growth and Problems of the City. University of
Zimbabwe, Harare, pp. 153–175.
Davies, R.J., 1981. The spatial formation of the South African city.
GeoJournal 2 (Supplementary Issue), 59–72.
Davies, R.J., 1996. Contemporary City Structuring: International
Geographical Insights. Society of South African Geographers,
Cape Town.
de Fuentes, G., Perez Medina, S., 1998. Segregation urbaine et
Ômodernisation’: Le cas de Merida, Yucatan. Annales de Geographie 107, 431–445.
de Swardt, W.R.P., 1970. Urbanism and Bantu––focus on town
planning and development in Soweto. In: Watts, H.L. (Ed.), Focus
on Cities. University of Natal, Institute of Social Research,
Durban, pp. 467–476.
Duncan, O.D., Duncan, B., 1955. A methodological analysis of
segregation indexes. American Sociological Review 20, 210–217.
Eichler, G., 1977. From colonialism to national independence: Algiers’
social ecology. GeoJournal 1 (5), 5–12.
Farley, R., Frey, W.H., 1994. Changes in the segregation of Whites
from Blacks during the 1980s: small steps toward a more integrated
society. American Sociological Review 59, 23–45.
Frey, W.H., Farley, R., 1996. Latino, Asian, and Black segregation in
U.S. metropolitan areas: are multiethnic metros different? Demography 33, 35–50.
Gervais-Lambony, P., Guillaume, P., 1999. Odi-Moretele: Ôurbanisation deplacee’ et reintegration territoriale en Afrique du Sud.
L’Espace Geographique 28, 159–169.
Gigaba, M., Maharaj, B., 1996. Land invasions during political
transition: the Wiggins saga in Cato Manor. Development Southern Africa 13, 215–235.
Greenberg, J.H., 1970. The Languages of Africa. Indiana University
Press, Bloomington.
Guillaume, P., 2001. Johannesburg: Geographies de l’exclusion.
Karthala, Paris.
156
A.J. Christopher / Geoforum 35 (2004) 145–156
Harrison, P., Todes, A., Watson, V., 1997. Transforming South
African cities: prospects for the economic development of urban
townships. Development Southern Africa 14, 43–60.
Hirschman, C., Alba, R., Farley, R., 2000. The meaning and
measurement of race in the U.S. census: glimpses into the future.
Demography 37, 381–393.
Hook, D., Vrdoljak, M., 2002. Gated communities, heterotopia and a
‘‘rights’’ of privilege: a Ôheterotopology’ of the South African
security park. Geoforum 33, 195–219.
Houssay-Holzschuch, M., 1999. Le Cap Ville-Sud-Africaine: Ville
blanche vies noires. L’Harmattan, Paris.
India, 1955. Report of the States Reorganization Commission.
Manager of Publications, New Delhi.
Jordan, A.C., 1973. Towards an African Literature: the Emergence of
Literary form in Xhosa. University of California Press, Berkeley.
Kantrowitz, N., 1969. Ethnic and racial segregation in the New York
metropolis. American Journal of Sociology 74, 447–491.
Kotze, N.J., Donaldson, S.E., 1998. Residential desegregation in two
South African cities: a comparative study of Bloemfontein and
Pietersburg. Urban Studies 35, 467–477.
Laitin, D.D., 1992. Language Repertoires and State Construction in
Africa. Cambridge University Press, Cambridge.
Lemon, A., 1991. Homes Apart: South Africa’s Segregated Cities. Paul
Chapman, London.
Maartens, J., 1998. Multilingualism and language policy in South
Africa. In: Extra, G., Maartens, J. (Eds.), Multilingualism in
a Multicultural Context. University of Tilburg Press, Tilburg,
pp. 15–36.
Magi, L.M., 1990. The diffusion of click sounds from Khoisan to
Bantu languages in Southern Africa. South African Geographer 17,
114–123.
Maharaj, B., 1999. The integrated community apartheid could not
destroy: the Warwick Avenue Triangle in Durban. Journal of
Southern African Studies 25, 249–266.
Massey, D.S., Denton, N.A., 1988. Dimensions of residential segregation. Social Forces 67, 281–315.
Massey, D.S., Denton, N.A., 1993. American Apartheid: Segregation
and the Making of the Underclass. Harvard University Press,
Cambridge, MA.
Mather, C., 1987. Residential segregation and Johannesburg’s Ôlocations in the sky’. South African Geographical Journal 68, 119–
128.
Mazrui, A.M., 2002. The English language in African education:
dependency and decolonization. In: Tollefson, J.W. (Ed.), Language Policies in Education: Critical Issues. Lawrence Erlbaum,
Mahwah, NJ, pp. 267–281.
McDermott, L., 1998. The case of English in South Africa. In: Extra,
G., Maartens, J. (Eds.), Multilingualism in a Multicultural Context. University of Tilburg Press, Tilburg, pp. 105–119.
Mohanty, S.P., Momin, A.R., 1996. Census as Social Document.
Rawat Publications, Jaipur.
Morris, P., 1980. Soweto: A Review of Existing Conditions and Some
Guidelines for Change. Urban Foundation, Johannesburg.
Moseley, C., Asher, R.E., 1994. Atlas of the World’s Languages.
Routledge, London.
Msimang, T., 1998. The nature and history of harmonisation of South
African languages. In: Prah, K.K. (Ed.), Between Distinction and
Extinction: the Harmonisation and Standardisation of African
Languages. Witwatersrand University Press, Johannesburg, pp.
165–172.
Oelofse, G., Dodson, B., 1997. Community, place and transformation:
a perceptual analysis of residents’ responses to an informal
settlement in South Africa. Geoforum 28, 91–101.
Oldfield, S., 2000. The centrality of community capacity in state lowincome housing provision in Cape Town, South Africa. Inter-
national Journal of Urban and Regional Research 24, 858–
872.
Ponelis, F.A., 1993. The development of Afrikaans. Peter Lang,
Frankfurt am Main.
Popke, E.J., 2000. Violence and memory in the reconstruction of South
Africa’s Cato Manor. Growth and Change 30, 235–254.
Pourtier, R., 2000. Brazzaville dans la guerre: crise urbaine et violences
politiques. Annales de Geographie 109, 3–20.
Prabhakaran, V., 1998. Indian languages in KwaZulu–Natal. In:
Extra, G., Maartens, J. (Eds.), Multilingualism in a Multicultural
Context. University of Tilburg Press, Tilburg, pp. 15–36.
Ratcliffe, P., 1996. Ethnicity in the 1991 Census––vol. 3 Social
Geography and Ethnicity in Britain: Geographical Space, Spatial
Concentration and Internal Migration. Her Majesty’s Stationary
Office, London.
Robinson, J., 1996. The Power of Apartheid: State, Power and Space
in South African cities. Butterworth-Heinemann, London.
Saff, G., 1996. Claiming a space in a changing South Africa: The
Ôsquatters’ of Marconi Beam, Cape Town. Annals of the Association of American Geographers 86, 235–255.
Smith, D.M., 1992. The Apartheid City and Beyond: Urbanization
and Social Change in South Africa. Routledge, London.
South Africa, 1954. Population Census 1946, Vol. IV––Languages
and literacy. Report no. UG18-1954. Government Printer, Pretoria.
South Africa, 1973. Population Census 1970. Report no. 02-05-01.
Government Printer, Pretoria.
South Africa, 1985. Population Census 1985: Unpublished language
tabulations. Central Statistical Services, Pretoria.
South Africa, 1992. Population Census 1991: Home Language. Report
no. 03-01-06 (1991). Central Statistical Services, Pretoria.
South Africa, 1999a. The People of South Africa: Population Census,
1996, Report no. 03-01-19 (1996). Statistics South Africa, Pretoria.
South Africa, 1999b. Annual Report 1999. Pan South African
Language Board, Pretoria.
South Africa, 2000. Language and language interaction in South
Africa: a national sociolinguistic survey. Pan South African
Language Board, Pretoria.
Taylor, R., Shaw, M., 1994. The Natal conflict. In: Brewer, J.
(Ed.), Restructuring South Africa. Routledge, London, pp. 35–52.
Telles, E.E., 1992. Residential segregation by skin color in Brazil.
American Sociological Review 57, 186–197.
Telles, E.E., 1995. Structural sources of socioeconomic segregation in
Brazilian metropolitan areas. American Journal of Sociology 100,
1199–1223.
Transkei, 1994. 1991 Population Census Report. Central Statistics
Office, Umtata.
van Bergen, M.M., Olivier, J.J., 1983. Sociale differensiasie tussen
Blanke etniese groepe in Pretoria. South African Geographer 11,
41–48.
van der Merwe, I.J., 1993. The urban geolinguistics of Cape Town.
GeoJournal 31, 409–417.
van der Merwe, I.J., 1995. Language change in South Africa: a
geographical perspective. GeoJournal 37, 513–523.
van der Merwe, I.J., van Niekerk, L.O., 1994. Language in South
Africa: Distribution and Change. University of Stellenbosch,
Department of Geography, Stellenbosch.
Venda, 1993. Population Census 1991. Office of the Chairman
of the Council for National Unity, Statistics Division, Thohoyandou.
Western, J., 1996. Outcast Cape Town. University of California Press,
Berkeley.
White, M.J., 1986. Segregation and diversity in population in
population distribution. Population Index 52, 198–221.
Williams, C.H., 1994. Called Unto Liberty! On Language and
Nationalism. Multilingual Matters, Clevedon.